For the past years, we have unfortunately seen several disastrous fires occurring in the forest, such as Amazon rainforest fires in Brazil and wildfires in western United States. The wildfires destroy millions of acres of forest (deforestation), which could have been used to preserve ecosystems, stock carbon dioxide and tackle climate change.
To address above problem, we aim to build the real-time monitoring system to predict the wildfires/climate change, with the help of satellite imagery combined with multi-spectral data (e.g., image) obtained from the unmanned aerial vehicle (UAV). Specifically, satellite imagery (provided by NASA via Global Forest Watch) will be applied to detect the long-term and large-scale evolution of forests and other possible causes of fires over space and time. However, it may lack of essential/fundamental measurements on the ground. To complement this, UAV will be sent to obtain the real-time and detailed information of a particular area, if uncertainties or potential issue appears from the satellite imagery (UAV data may be obtained from online database, and/or form partners and/or from real experiment). Innovative Artificial Intelligence (AI) model (e.g., deep neural networks or DNN) along with novel data fusion technologies will be proposed to combine satellite imagery and data from UAV, to make the real-time prediction and determine the possibility of forest fire. The newly developed solution can help the relevant authorities to identify potential risks and increase their response time. This project can also be extended to track wildlife, monitor natural resource, prevent illegal activity and climate change, by extracting other key features from the satellite imagery combined with UAV’s data, with the newly developed approach.
The PhD student will work with a multidisciplinary supervision team and receive specific research training on environmental informatics, artificial intelligence, image processing, and enviromnet modelling/prediction.